Rugby union is a high-intensity intermittent team sport of 80 min duration. It is characterised by its combative nature, requiring aggressive and forceful plays in many areas of the game.1 Whilst research has quantified the physical work performed during a game by monitoring heart rate and blood lactate, and performing movement analyses,2^–4 there is a lack of research examining biochemical changes that occur during rugby union games.1

The monitoring of biochemical changes during exercise is a valid way of understanding the physiological stress placed upon an athlete.5 Physical activity has been shown to influence biochemical markers according to the activity’s intensity, duration and mode of exercise.6^–9 Numerous studies have been completed under controlled laboratory conditions, but the possible influence of the competitive environment dictates that the responses should be measured during competition.

Muscle soreness and damage from intense exercise, which is commonly reported by players in the days after a game, is associated with a loss of muscle force generating capacity.10 Common indicators of muscle damage are the presence of muscle proteins and enzymes in the blood, such as creatine kinase (CK).11 12 Research examining the factors influencing muscle soreness and damage has shown that eccentric muscular work is the major contributor to increases in creatine kinase concentration ([CK]), which peaks 2−3 days after a damaging exercise bout.13 14 However, increasing evidence suggests that a greater [CK] may result from physical impact or blunt trauma.1 11 15 16 A recent study by Takarada1 showed that the peak [CK] of well-trained amateur rugby players was highly correlated to the number of tackles the player was involved in during a game. With limited research examining the relationship between physical impact and the [CK] response, and more importantly with respect to rugby union, the association reported by Takarada has not been verified. In this sense, it is still not known whether [CK] can be used as an indication of the amount of intense exercise bouts performed by a player, or whether it is more representative of the number and intensity of impacts. This is important for those who wish to use [CK] to better understand the physical demands placed upon a player in order to better prescribe recovery interventions. Therefore, the primary purpose of this study was to investigate the relationship between the pre- to post-game changes in creatine kinase concentration (Δ[CK]) and impact-related game statistics in elite rugby union players.

METHODS

Participants

Twenty-three elite male rugby union players (mean (SD) age 25 (3) years, height 184 (9) cm, body mass 99.2 (10.1) kg, sum of eight skinfolds 87.7 (22.4) mm) volunteered to participate in the study. The sample was drawn from a squad of players selected to play in the New Zealand National Provincial Championship (NPC). The average training commitment of the subjects was 12 (SD 2) hours per week. Training consisted of resistance training, general conditioning, and speed, skill and team trainings. Owing to the highly trained nature of the sample from which the subjects were drawn, 22 players were selected to play each week. Each subject provided written informed consent and the project was approved by The Waikato Institute of Technology’s Ethics Committee.

Research design

The primary research design used was a single-group, repeated-measures pre−post game design. Each game was separated by at least 5 days, which was dependent upon the competition draw, and normal training continued between games. Transdermal exudate samples were obtained for CK during home games in the NPC season, in which five regular season games were observed. During each game, the subjects provided samples at two different points; pre-game (approximately 210 min prior to the start of each individual game) and post-game (as soon as possible after the completion of the event, within a maximum time window of 30 min). These sampling times were influenced by the availability of the players before and after the games; thus, disturbances to the subjects prior to each game was minimal. Sampling was performed using a previously described method.17 18

Transdermal sampling and analysis

The electrosonophoretic devise (ESoP) was used to collect transdermal samples. The ESoP uses a combination of ultrasound (frequency of 1 MHz and generator power of 0.35 W/cm²) and electric current (9 V) to augment the movements of analytes to the skin surface. A mixture of standard ultrasound gel (Aquasonic, Parker, USA) containing 1% wt/vol of emu oil extract (Emea, New Zealand), 1% wt/vol sodium lauryl sulphate and 1% wt/vol ethanol was applied to the skin at the sampling site (ventral forearm) prior to ultrasound application. Previous experimental work has demonstrated this mixture facilitates transdermal movement of interstitial fluid.17 19 Collection fluid was circulated through the sampling head at 1.2 ml/min, with samples collected in a 1.7 ml-graduated microtube (1210–04, Invitrogen, Auckland, New Zealand). All samples were stored at −20°C until analysed by assay.18

Half of each sample was concentrated 10-fold, and the other half 100-fold by vacuum extraction and nitrogen blow-off before being assayed for [CK] using a derivation of the Rosalki method,20 with a Du Series 500 spectrophotometer (Beckman Coulter, Pty. Ltd., Australia). If a high enough sensitivity was not achieved, samples were subjected to volatisation and gas chromatographic mass spectrometry. The mean coefficient of variation (CV) for CK assays was 6.4 (3.1)%. Reliability and validity studies have shown ESoP samples to be highly correlated with plasma at rest, as well as during exercise and recovery.17

Tackle and game statistics

Game statistics and video for analysis were drawn from the AnalyRugby computer software package (AnalySports, Version AS10.0307, 2002, Palmerston North, NZ). The identified statistics included game time, time defending, tackles made, hit-ups, first three players on attack (being one of the first three players at the breakdown while their team is attacking), first three players on defence (being one of the first three players at the breakdown while their team is defending) and the number of scrums. In addition to these game statistics, total impact (sum of tackles made, hit-ups, first three on attack and first three on defence) and collisions per minute (impact/game time) were also calculated. These specific game statistics were deemed to be important in determining the relationship between impact and [CK] and were calculated for each individual player across each of the five games.

Statistical methods

The data were analysed as a whole group as well as separated into two groups that were representative of the players’ positions on the field (ie, forwards vs backs); there were insufficient data to analyse smaller player subgroups (eg, inside vs outside backs or front row vs loose forwards). Means and standard deviations were calculated for the Δ[CK] and identified game statistics. Given that game statistics and CK samples were taken across several games for each player, standard correlation and regression techniques could not be used. Thus, relationships between tackle statistics and Δ[CK] were quantified using the weighted between-subject Pearson product moment correlation as per the methods of Bland and Altman.21 A backwards random-effects maximum likelihood regression was performed on the game statistics to develop a predictive equation for the Δ[CK]. All statistical analysis was performed using Microsoft Excel (2002, version 10.4302.4219 SP-2; Microsoft Corporation, USA) and Stata (version 8, Stata Corporation, TX, USA), with statistical significance set at p<0.05.

RESULTS

Each subject participated in an average of 3.4 games (ie, an average of 3.4 data points were obtained for each subject). The average Δ[CK] (mean (SD)) during a game was 926.8 (204.2) IU. Compared to the backs, the forwards displayed a significantly higher Δ[CK] (Δbacks  = 545.4 (340.6); Δforwards  = 1439.5 (676.9)). Game time and time defending were the only game statistics that were not significantly different between backs and forwards (table 1). However, both game time and time defending were significantly correlated with the Δ[CK] of both the forwards and backs (table 2). The correlation between the other game statistics and Δ[CK] appeared variable between the backs and forwards. For example, scrum number was significantly correlated to Δ[CK] in the forwards (r = 0.73), while hit-ups and first three on attack were significantly correlated to Δ[CK] in the backs (r = 0.74 and r = 0.79, respectively).

Backwards random-effects maximum likelihood regression equations were developed for the forwards and backs:

Δ[CK] FORWARDS  = 1759 + (47.52 × first three on attack) + (35.95× tackles made) + (−2892.54× collisions per min)

Δ[CK] BACKS  = 31.91 + (92.25× impact) + (12.77 × time defending) + (−65.13× first three on attack) + (−117.42× tackles made)

The mean of each independent variable (game statistics) obtained during the study was entered into the regression equations to produce a predicted mean. The predicted Δ[CK] (95% confidence limit) was 1439.8 (204.9) IU for the forwards and 545.3 (78.0) IU for the backs. The predicted Δ[CK] was significantly correlated with the actual Δ[CK] for forwards (r = 0.69; p<0.01) and backs (r = 0.74; p<0.01). The prediction equations for both positional groups used tackles made and first three on attack as contributing variables; however, the contribution differed between groups. All three game statistics that were significantly correlated to the forwards Δ[CK] (game time, time defending and scrum number), and two out of the four significantly correlated game statistics to the backs Δ[CK] (game time and hit ups) were not required for the accurate prediction of Δ[CK].

DISCUSSION

The primary purpose of this study was to investigate the relationship between the pre-game to post-game changes in creatine kinase concentration (Δ [CK]) and impact-related game statistics in elite rugby union players. Creatine kinase concentration increased significantly from pre-game to post-game. During acute muscle damage, CK is released into the interstitial fluid before being transported back to the blood through the lymphatic system, thus causing an increase in serum [CK].11 The significant increase in [CK] from pre-game to post-game indicates that a competitive rugby union game induces a significant level of muscular damage. The magnitude of Δ[CK] observed in the present study is greater than that reported in Japanese rugby union players.1 The difference may be due to different tactics or game patterns of Japanese rugby compared to New Zealand rugby. Differences in body size between the two study samples (the current sample were an average ∼5 cm taller and ∼12 kg heavier in body mass) might also affect the magnitude of impact and its effect on the players.

The Δ[CK] from pre-game to post-game was significantly greater in the forwards compared to backs. Given that both forwards and backs perform a significant amount of running in a game, we hypothesised that differences in [CK] between forwards and backs might be related to the number of physical contacts in a game. A number of previous studies have shown a significant influence of impacts on the Δ[CK]. Zuliani et al16 reported an elevation in [CK] after a regular boxing match, but not after shadow boxing that did not involve direct punches by an opponent. Additionally, studies that have explored the [CK] response in American football and rugby union players have shown an augmented effect of blunt trauma incurred during tackling upon [CK].1 11 15 22 As the forwards perform a higher number of tasks involving collisions in a game (table 1), it is very likely that differences in the Δ[CK] between backs and forwards can be attributed to the differing number of impacts incurred during the game. In order to determine which impact-related factors influence the [CK] difference between backs and forwards, regression equations were formulated to predict Δ[CK]. As hypothesised, the prediction of Δ[CK] required the incorporation of different impact statistics for forwards and backs. Furthermore, the moderate confidence limits of the predicted values from the regression equation (95% confidence limit; forwards  =  204.9 IU; backs  =  78.0 IU) and the strong correlation between actual and predicted Δ[CK] values indicate that game statistics can be used as reasonable indicators of the magnitude of the Δ[CK].

The regression equation for the forwards indicates that a significant proportion of the Δ[CK] can be explained by the tackles made (increase of 35.95 per tackle) and first three on attack (increase of 47.52 per event). This is unsurprising as it is well known that forwards perform a high number of tackles and compete in rucks and mauls frequently.4 23 The contribution of these particular contact areas to the prediction equation may be due to an equal involvement of the entire forward pack in these tasks. The fact that being one of the first three players to a ruck or maul in attack, but not defence, greatly impacts on Δ[CK] might indicate that these collisions are of greater magnitude or are performed with a technique that incurs greater muscle damage. A better understanding of ruck and maul technique might allow a revision that could reduce the effects of these impacts. Interestingly, scrum number was significantly correlated to Δ[CK] within the forwards, but was not included in the regression equation. The scrum produces high-impact forces when the two packs engage (approximately half a ton); almost twice as much force than that produced during a sustained push.24 25 Therefore, when the momentum from the two packs that produce similar forces are combined, the resulting impact could create substantial physical trauma and impact-related muscular damage to the front row. One reason for its omission from the regression equations is that it was possibly highly correlated with other statistics that were also included. Another reason might be that, since the loose forwards largely act in a supporting role in the scrum, the impact at engagement may have a negligible effect on them and their [CK] increase.4 24 25 By comparison, all forwards are involved in tackling and ruck and maul situations, which might be one reason why these variables have greater influence on the ability to predict forwards' Δ[CK] as an entire group.

The major positive contributing factor to the predicted Δ[CK] in backs was the impact measure (the sum of tackles made, hit ups, first three on attack and first three on defence). In contrast, tackles made and first three on attack have a negative contribution within the regression equation. While tackles made and first three on attack make up approximately 80% of the total impact measure in this study, these types of events in isolation do not appear to contribute to the increase in [CK] pre-game to post-game. One reason is that whereas the number of impacts in general might increase the Δ[CK] in backs, tackling and ruck and maul techniques of the backs might induce less damage. Alternatively, their performance by a back might stop them being involved in subsequent phases of play, and allow their recovery from the impacts. Indeed, the backs are involved in longer distance running but less tackling than the forwards, therefore resulting in a greater time between tackles,4 23 and the majority of time spent by a back during games is either performed walking or jogging, thus acting as a passive type of recovery and potentially enhancing the CK clearance.4 18 However, more research is required to more completely understand the effects of the types of impact on Δ[CK].

There are several differences in the types of events found to influence Δ[CK] in this study compared to that of Takarada,1 who investigated [CK] responses in a similar rugby union player sample. The current project used 23 subjects over a period of five games (approximately 115 individual measures), whereas Takarada used 15 subjects within one game. Although the correlation reported by Takarada between impact (tackles) and [CK] was very high (r = 0.922; p<0.01), the statistical power compared to the present study is much less due to the difference in sample size.26 Furthermore, Takarada only investigated the relationship of tackles to the [CK]. The tackle variable was defined as “the total number of times that a player tackled or was tackled from in front”.[1; p417] Therefore, Takarada ignored impacts from other directions and other variables such as game time, first three in attack and the scrums within the forwards, which were found to be highly correlated to the Δ[CK] in the present study.

Another difference between this study and others was the sampling technique. Previous research has measured [CK] from blood, whereas the present study used the non-invasive ESoP technique.1 11 15 18 22 Electrosonophoresis draws samples from interstitial fluid, so the [CK] within interstitial fluid may not represent the [CK] within the blood. Interstitial fluid is not a direct measure of blood levels of analytes; nor do all analytes freely partition between these different body fluids.17 However, interstitial fluid correlates strongly with, and is driven by, blood levels and can therefore be used to estimate changes in blood concentrations.17 Thus, while the values might differ slightly, the use of ESoP has been shown to be both valid and reliable.

CONCLUSIONS

Creatine kinase concentration increased significantly from pre-game to post-game in a position-specific manner in elite rugby union players. The game statistics showing a strong correlation with Δ[CK] and the regression equations derived in order to predict the Δ[CK] varied between positions, which probably reflect their different playing roles. The significant correlation between predicted Δ[CK] derived from the regression equations and actual Δ[CK] (r = 0.69 and r = 0.74 for the forwards and backs, respectively) and the moderate confidence limits of the regression predictions are indicative of a strong ability to predict the Δ[CK] from pre-game to post-game. Given that [CK] is an indicator of muscle damage, the prediction of Δ[CK] by regression equations provides an indication of the degree of muscular damage incurred by a particular player through a rugby union game. Such information may be used to manage workload and training intensity of individuals within the team to reduce the effect of delayed onset of muscular soreness (DOMS) and possible soft tissue injury.